During the last few decades, many global tide gauges have increased their sampling resolution, which enabled the research of short-period sea-level oscillations and improved assessments of sea-level extremes. Apart from seismic tsunamis, atmospherically-driven short-period oscillations and their extreme manifestations known as meteorological tsunamis are shown to contribute to extreme sea levels. Meteorological tsunamis are sea-level oscillations which occur on periods of seismic tsunamis (from a few minutes to a few hours) and are generated by intensive propagating atmospheric disturbances. In this thesis, available global 1-minute sea-level observations are merged and unified, then subjected to quality control procedures and finally filtered to extract their high-frequency signal (with period T < 2h). The dataset named MISELA (MInute SEa - Level Analysis data) has been created containing observations from 331 tide gauges worldwide. Global and regional statistical analyses have been conducted using the MISELA dataset and a climatology of atmospherically-induced high-frequency sea-level oscillations has been constructed. The connection between high-frequency oscillations and relevant synoptic variables has been quantified. Moreover, specific synoptic patterns have been recognized during meteorological tsunamis in many world regions. Along with global analyses of high-frequency signal, this thesis studies the phenomenon marrobbio, a meteorological tsunami which occurs on the coasts of Sicily and Lampedusa. The observations from 2007 have been analysed and six marrobbio events with a range of oscillations higher than 40 cm have been detected. The oscillations are shown to have the highest range in Mazara del Vallo, a known “hotspot“ of marrobbio, located at the southwest coast of Sicily.